The present invention relates to an apparatus for manufacturing a thin-film solar cell formed of a compound semiconductor which consists of group IB elements, group IIIB elements and group VIB elements in the periodic table, and a method of manufacturing such a thin-film solar cell.
Attention has recently been focused on a compound semiconductor consisting of group IB, IIIB, and VIB elements in the periodic table which makes it possible to manufacture a thin-film solar cell having a superior photoelectric conversion efficiency and a large area at a low cost. Particularly, CuInSe.sub.2 has the following advantageous characteristics:
(1) an absorption coefficient .alpha. as high as about 10.sup.5 /cm which allows sufficient absorption of solar light even when the solar cell is formed as thin as about 2 .mu.m, PA1 (2) a forbidden band of 1.1 eV which is suitable for photoelectric conversion of solar light, and PA1 (3) a considerably smaller degree of photo-deterioration compared to amorphous silicon. For these reasons, the compound semiconductor has been receiving the most attention. A method of manufacturing a thin-film solar cell is disclosed in, for example, Unexamined Japanese Patent Publication No. Sho. 61-237476.
FIG. 5 is a cross-sectional view of a conventional apparatus for manufacturing a thin-film solar cell.
This manufacturing apparatus constitutes a part of a heat treatment furnace disclosed in Unexamined Japanese Patent Publication No. Sho. 61-237476. A narrow and long cylindrical core pipe 51 is provided with a baffle 52, and this baffle is circumferentially disposed along the internal wall of the core pipe at the longitudinal center thereof. As a result, the core pipe 51 is divided into two chambers 53 and 54.
A thin-film solar cell 55 to be subjected to heat treatment comprises a glass substrate, an Mo film as a conductive film laid on the glass substrate, and a Cu--In/Se film laid on the Mo film. At this time, the Cu--In/Se film is not alloyed yet. A plurality of thin-film solar cells 55 to be subjected to heat treatment are placed in the chamber 53 (the chamber positioned on the left side in the drawing), and a crucible 56 which contains a Se material is placed in the other chamber 54 (the chamber positioned on the right side in the drawing). The core pipe 51 is heated to heat the thin-film solar cells 55 and the crucible 56, as a result of which Se is gasified. As shown in the drawing, an Ar gas is supplied into the core pipe 51 from the right side, so that the Se gas flows into the chamber 53 (the chamber positioned on the left side in the drawing). The solar cells 55 are then thermally treated, whereby the Cu--In/Se film of the solar cell is transformed into a CuInSe.sub.2 alloy film.
Several propositions have already been made with respect to the relationship between temperature and duration of the heat treatment. For example, as shown in FIG. 6, the temperature of the chamber is increased from room temperature to 200-250 centigrades at a rate of 30 centigrade/min. The thus increased temperature is maintained for about 30 to 60 minutes, and the temperature is further increased to 400-450 centigrades at a rate of 30 centigrade/min. The thus increased temperature is then maintained for about 2 to 4 hours. The temperature is then cooled to room temperature.
The above described conventional heat treatment furnace has a simple structure and requires an inexpensive installation cost, and therefore it is widely used. However, the heat treatment furnace of this type is an open furnace, and the Ar and Se gases are kept flowing in the furnace. This results in considerably large amounts of the Se and Ar gases being used, which in turn adds to the cost. Further, this heat treatment furnace is not closed, and hence it is impossible to increase the pressure of the Se gas. The Se gas merely flows above and below the thin-film solar cell 55 or between the thin-film solar cells. For this reason, the Se gas fails to sufficiently react with the thin-film solar cell 55, which makes it difficult to form the CuInSe.sub.2 alloy film. Eventually, a production yield drops.
Also, Unexamined Japanese Patent Publication No. Sho-61-237476 discloses a method of manufacturing the ternary alloy (CuInSe.sub.2). According to this method, a precursor is formed by electrically depositing copper and indium on a conductive substrate on top of each other, and the thus formed precursor is heated in the flow of an inert gas which includes hydrogen seleniumide, whereby the copper-indium-selenium ternary alloy (CuInSe.sub.2) layer is formed.
However, the method has such a problem that the conductive substrate and the ternary alloy layer are poorly in contact with each other and, therefore, the ternary alloy layer is apt to delaminate. Eventually, the thus manufactured solar cell has inferior characteristics.
Further, the precursor is heated in the flow of the inert gas including hydrogen seleniumide, and hence several tens to several hundreds as much hydrogen seleniumide as the stoichiometrically required volume of hydrogen selenide are required. The hydrogen selenide is significantly toxic, and therefore it is really troublesome to handle. Still further, the use of expensive inert gas adds to the cost.
There is other known method of manufacturing the copper-indium-selenium ternary alloy (CuInSe.sub.2) as disclosed in International Publication Number WO 92/05586. A layer consisting of three components (hereinafter referred to as a precursor layer) is formed by plating, and the thus formed precursor layer is subjected to heat treatment, whereby a ternary alloy layer consisting of three components is formed.
It is not easy to form the precursor layer by controlling the ratio of the three components. Particularly, in the case of the technique as disclosed in Internal Publication Number 92/05586, a selenium powder is dispersed into a plating solution, and therefore it is necessary to continuously stir the solution during the course of plating. The stirring of the solution exerts a considerable influence on the composition of the plating layer.
The ratio of these three components is apt to change in the thicknesswise direction of the plating layer (the ratio of copper atoms in the plating layer is large when the plating is started, and it subsequently becomes smaller while the ratio of indium atoms becomes larger). Non-destructive analyzing means which analyze the real ratio of the atoms has not been available.
Therefore, it is very difficult to manufacture a thin-film solar cell comprising the three components at a stably controlled ratio.
Further, according to the above described conventional technique, a stable copper-indium alloy develops during the course of the plating operation. This copper-indium alloy prevents the growth of a copper-indium-selenium alloy crystal which is generated when the plating layer undergoes heat treatment after the plating operation. The copper-indium alloy also hinders the orientation of the copper-indium-selenium alloy crystal. The thus formed absorbing layer comprises the residual copper-indium alloy which does not have photoelectric converting characteristics, and hence the overall photoelectric conversion efficiency becomes low.
As well as the above described method, it is conceivable to apply sputtering and PVD methods, in which a layer is formed in a depressurized state, when forming the precursor layer. However, these methods provide lower productivity and also require large installation costs.